针对精确估计信号频率问题,对经典全相位时移相位差估计器做了估计性能改进.通过深入剖析研究卷积窗类型和频率偏离值对全相位 FFT 峰值谱幅度的影响,进而提出两项改进措施:在全相位数据处理环节引入无窗卷积窗,在原有估计器的输出引入入频移补偿. 使得不管频率怎样偏离,峰值谱总能接近理想幅值,从而既提高了估计器抵御噪声的能力,又显著提高了估计器的精度.实验结果表明,该方法的改进估计器,其精度高于原有的全相位时移相位差法和改进的 Candan 估计器,且对于任意频偏情况,其频率估计方差均紧靠克拉美罗限,具有较广泛的应用前景.
For the issue of accurately estimating a signal’s frequency,an essential innovation on the classical all-phase time-shift phase difference frequency estimator was proposed. By exploring the mechanism of how the con-volution window and the frequency offset affect the magnitude of all-phase FFT peak spectral bin,two improvement measures were put forth:A no-window convolution window was incorporated in all-phase data preprocessing;Techniques of frequency offset and compensation were applied to the output of original estimator. These two measures ensure that the peak spectral bin’s magnitude approximates the ideal amplitude in any case of frequency offset,thereby enhancing the robustness to noise and the estimator’s accuracy significantly. Simulation results showthat,no matter how the frequency offset varies,the proposed estimator possesses a higher accuracy than the original all-phase time-shift phase difference frequency estimator and the improved Candan estimator. Furthermore,the frequency es-timate variance of the proposed estimator approximates the Cramer-Rao lower bound for any case of frequency offset,which presents vast potential for future development.